Early life circumstances and their impact on menarche and menopause

Abstract

Ages at menarche and menopause have been shown to be associated with adverse health outcomes in later life. For example, earlier menarche and later menopause have been independently linked to higher risk of breast cancer. Earlier menarche may also be associated with an increased risk of endometrial cancer, menstrual problems and adult obesity. Given the associations of ages at menarche and menopause with future health outcomes, it is important to establish what factors across life, and generations, may influence these. This article examines the associations of early life factors, namely birthweight, bodyweight and growth during childhood, childhood socioeconomic circumstances and psychosocial factors with ages at menarche and menopause. It examines possible explanations of the associations found, including life history theory, and discusses areas for future research.

Reproductive health, from menarche to menopause, is not only understood as being integral to women’s overall health and wellbeing, but is increasingly recognized as a sentinel of chronic disease in later life [1-3]. For example, both earlier menarche and later menopause are independently associated with higher risk of breast cancer [4,5], and may also be associated with an increased risk of endometrial cancer [6-8]. When examined together, the length of time between menarche and menopause, which provides a crude indicator of lifetime estrogen exposure, has also been shown to be associated with these outcomes [9]. Earlier menarche may also be a risk factor for adult obesity [10]. By contrast, age at menarche has little bearing on cardiovascular risk, and only premature age at menopause is associated with significantly higher risk of heart disease [1,11].

There is growing evidence that the physical and social environment in previous generations, and from preconception to midlife, influence both reproductive health [1] and, later, chronic disease [12,13]. Menarche and menopause are the two major components of women’s reproductive lives, since the interval between them determines the natural reproductive phase [14]. Therefore, the aim of this article is to update information on the role that early life factors play on these key aspects of reproductive health, namely age at menarche and the timing of menopause.

Methods

A comprehensive search of the literature published in the English language between 1980 and July 2008 was performed using the terms ‘childhood, early life, fetal, in utero or epigenetic’ and ‘age at menarche, age at menopause or hysterectomy’. Of the studies identified, we have chosen to include those with a sample size of at least 400, as it has been suggested that these will have sufficient power to study menopause [15]. We assumed this sample size was also sufficient for the study of menarche since this characteristic (excluding precocious menarche) is less variable than age at menopause [16].

Age of menarche

Menarche marks the commencement of the reproductive phase of a woman’s life. Records suggest that there was a secular decline in the average age at menarche in developed countries across the 19th and 20th Centuries until the 1950s [17], with an average age at menarche of 15-17 years reported in the mid-19th Century [18,19]. It has been suggested that pubertal timing may then have plateaued in the 1950s, although further small declines have subsequently been reported [17]. While recent estimates of the median age of menarche vary between 13 to 16 years, recent data suggest that the age of 15 years represents the 95-98th percentile for menarche [17]. A review of family and twin studies has highlighted the key role of genetic factors in determining the timing of menarche, with heritability estimates ranging from 0.44 to 0.72 [20,21]. In examining early life predictors of age at menarche, three major themes have emerged, namely the effects of body size, social circumstances and exposure to unfavorable psychological circumstances. The effects of these environmental influences on the timing of menarche can be explained, from an evolutionary-development perspective, in terms of life history theories.

Life history theory perspectives on the timing of menarche

The main life history theories have been elaborated by Ellis [22] as the energetics theory, and the four psychosocial models of pubertal timing: psychosocial acceleration, parental investment, stress suppression and child development theory. However, the theories differ in their conceptualization of the nature, extent and direction of environmental influences on the age of menarche, and the effect that the timing of menarche has on subsequent reproductive characteristics [22].

The energetics theory suggests that energy availability during childhood influences the timing of menarche. It hypothesizes that girls who were exposed to a chronically poor nutritional environment will grow more slowly, experience later pubertal development (relative to their genetic potential), and reach relatively small adult size compared with those children who were exposed to greater food availability. This is consistent with an earlier idea regarding the impact of relative fatness that follows from the close relationship of average critical bodyweight on age at menarche [23]. The psychosocial acceleration theory posits that the experience of high levels of emotional stress in and around a girl’s family leads to earlier menarche in order that she can maximize her chance of leaving descendents [24,25]. Based on the same logic, the parental-investment theory hypothesizes a special role for the father and other men in influencing the timing of menarche. The stress-suppression theory [26] proposes that early adversity, whether it is by adverse physical or social conditions or psychosocial stress, causes a delay in pubertal development until better times. Last, Ellis describes child-development theory as reconceptualizing ‘the age at menarche as the end point of a developmental strategy that conditionally alters the length of childhood in response to the composition and quality of family environments’ [22].

Birthweight & childhood growth

Several studies have examined birth size and infant growth in relation to age at menarche (Table 1) [27-31]. In a study investigating Filipino girls, long and thin infants had earlier menarche and this effect was strongest for girls with higher growth rates from 0-6 months [31]. The Medical Research Council (MRC) National Survey of Health and Development (NSHD) found, in a group of 2547 British girls, that rapid growth in infancy was associated with an earlier menarche. After adjusting for growth in infancy, higher birthweight was also predictive of an earlier menarche [29]; however, adjustment for growth later in childhood attenuated both of these associations. In another study, Swiss girls who were small for gestational age had earlier menarche than those of normal size for gestational age [28]. As in the previous study, the results were no longer significant when adjusted for childhood growth. A Polish study also found that girls born small for gestational age were more likely to have reached menarche by the age of 14 years, compared with other girls, but the authors were unable to adjust for childhood growth [32]. These results suggest that while age at menarche may be determined, in part, by factors in utero or in infancy, perhaps by programmed release of gonadotrophin [29], such effects are mediated through early childhood growth.

Table 1.

Early life factors and their associations with age at menarche.

Study	Study population	Country of study	n	Early life factor(s) examined	Association(s) with age at menarche	Ref.
Adadevoh et al. (1989)	Cross-sectional study of school girls	Ghana	2087	Father’s occupation and place of residence	Higher father’s occupational class and living in an urban area → earlier menarche	[41]
Attallah et al. (1983)	Cross-sectional study of school girls	Sudan	1372	Family income group	Higher family income → earlier menarche	[45]
Bielicki et al. (1986)	Cross-sectional study of girls from elementary schools	Poland	20373	Parental education and father’s occupation	Higher levels of parental education and higher paternal occupational class → earlier menarche	[48]
Billewicz et al. (1981)	Girls taking part in a longitudinal study of growth and development in Newcastle upon Tyne	UK	699	Birth order, number of siblings, birthweight, weight and height at 5 years of age, age at peak height velocity, arm circumference, subscapular skinfold, triceps skinfold and father’s occupation	Larger family → later menarche Father in nonmanual occupation and higher weight and height at the age of 5 years → earlier menarche Birthweight → no association	[46]
Bogaert (2005)	A national probability sample	USA	1921	Presence of parents at the age of 14 years	Father absent → earlier menarche Mother absent and stepfather present → no association	[55]
Bogaert (2008)	National Survey of Sexual Attitudes and Lifestyles (NATSAL, 2000) respondents	UK	5913	Presence of parents up to age 16 years	Father absent → earlier menarche Mother absent → no association	[56]
Brown et al. (2004)	Children in the community study	USA	401	Childhood abuse and neglect	Two or more episodes of sexual abuse → earlier menarche	[52]
Chavarro et al. (2004)	Women entering an undergraduate program at the National University of Colombia	Colombia	3206	Place of birth, childhood migration, parental education, socioeconomic background, family size and physical activity in year before menarche	Greater family size and increased hours of physical activity → later menarche Urban birth place and higher socioeconomic position → earlier menarche	[42]
Cooper et al. (1996)	Medical Research Council National Survey of Health and Development	UK	1471	Birthweight, height and weight at the age of 7 years	Higher birthweight and lower weight at 7 years of age → later menarche	[30]
Ersoy et al. (2004)	Cross-sectional study of school girls	Turkey	1000	Paternal education and occupation	No association	[139]
Farid-Coupal et al. (1981)	National Human Growth, Nutrition and Family Survey	Venezuela	955	Socioeconomic position	Higher socioeconomic position → earlier menarche	[140]
Jorm et al. (2004)	PATH Through Life Project	Australia	3702	Childhood adversity scale (which consisted of measures of lack of parental affection, nervous or emotional trouble or depression of parents, parental alcohol abuse and ten types of parental mistreatment)	Higher levels of childhood adversity → earlier menarche	[53]
Junqueira et al. (2003)	Pro-Saude Study	Brazil	2217	Father’s education level	Lower paternal education levels → greater declines in average age at menarche over time	[141]
Khan et al. (1996)	Women who participated in a nutrition intervention study as children	Guatemala	497	Height-for-age at the age of 3 years, dietary supplementation from birth to the age of 7 years, average energy intake from birth to the age of 3 years, skeletal maturation, percentage of time ill with diarrhea or respiratory illness up to the age of 3 years	Higher degree of stunting and higher percentage of time ill → later menarche Increased caloric intake and height-for-age → earlier menarche	[142]
Łaska-Mierzejewska et al. (1982)	Group of school girls participating in a cross-sectional survey	Poland	5546	Father’s education level and family size	Higher paternal education level → earlier menarche Larger family size → later menarche	[51]
Lumey et al. (1997)	Women born in a specific hospital between 1944 and 1946	The Netherlands	700	Exposure to Dutch famine by trimester	Trimester of exposure to famine → no association	[143]
Marrodan et al. (1999)	Cross-sectional study of school girls	Spain	811	Area of residence (urban vs rural)	No association	[49]
Mendleet al. (2006)	Children of twins study	Australia	1284	Parental divorce and presence of parents	Presence of stepfather or stepuncle → earlier menarche Parental divorce → no association	[55]
Moisan et al. (1990)	Nested case-control study of school girls	Canada	666	Diet, weight, height, abdominal skinfold, suprailiac skinfold, physical activity and energy expenditure	Greater height, weight and physical activity levels → earlier menarche Weak positive association between energy intake, energy expenditure and age at menarche	[144]
Oduntan et al. (1976)	Cross-sectional study of school girls	Nigeria	2357	Parental education, father’s occupation, area of residence, number of siblings, birth order and patterns of marriage in family	Urban residence, higher parental education levels and higher occupational class of father → earlier menarche Number of siblings, birth order and patterns of marriage → No association	[40]
Pasquet et al. (1999)	Participants in various cross-sectional surveys	Cameroon	911	Urban or rural location	Urban location → earlier menarche	[43]
Pesonen et al. (2008)	Helsinki Birth Cohort	Finland	511	Evacuation without parents during the war	Experiencing evacuation → earlier menarche	[59]
Prebeg et al. (2000)	Women participating in one of three cross-sectional surveys undertaken in 1981, 1985 and 1996	Croatia	3607	Family socionecomic position and number of siblings 1996 survey only: relocation, death of family member or damage of home or property during Balkan war	Number of siblings → no association In 1996 survey: family member killed; or home or property: - Damaged → later menarche - Relocation → no association	[50]
Rao et al. (1998)	Longitudinal study of school girls	India	533	Socioeconomic position, peak height velocity and peak weight velocity	Higher socioeconomic position → earlier menarche Time from peak height and weight velocity to menarche did not differ by socioeconomic position	[27]
Romans et al. (2003)	The Otago Women’s Health Child Abuse Survey	New Zealand	488	Family socioeconomic position, presence of parents, family conflict, quality of relationship with parents, physical and sexual abuse	Univariate analyses: lower socioeconomic position; father absent; family conflict; poor relationships with parents; abuse → earlier menarche In fully adjusted analyses: longer duration of abuse → earlier menarche	[57]
Schooling et al. (2008)	Guangzhou Biobank Cohort Study	China	7273	Socioeconomic position (father’s occupation and women’s education as a marker of family wealth in childhood)	Lower socioeconomic position → earlier menarche	[38]
Simondon et al. (1997)	Cross-sectional study of girls born between 1978 and 1984	Senegal	1181	Seasonal work migration	No association	[145]
Sloboda et al. (2007)	Western Australia Pregnancy (Raine) Cohort Study	Australia	776	Expected birthweight ratio and BMI at the age of 8 years	Expected birthweight ratio lower than median → earlier menarche Higher BMI at age 8 years → earlier menarche	[34]
Tahirović (1998)	Participants in a cross-sectional study who were deported from Srebreica to refugee camps in Tuzla or who lived in unoccupied areas of Bosnia and Herzegovina	Bosnia	5820	Relocation to a refugee camp	Relocation to refugee camp → later menarche	[65]
Ulijaszek et al. (1991)	Girls aged 9-17 years who were of European, Afro-Caribbean or Indo-Pakistani descent and living in designated neighbourhoods in London	UK	2177	Father’s occupational class, birth order and family size	Larger family size → later menarche (Indo-Pakistani and European girls) Higher social class size → earlier menarche (in Afro-Caribbean and European girls)	[47]
van Noord et al. (1991)	DOM project, participants in a breast cancer screening program	The Netherlands	16,583	Exposure to World War II and the Dutch famine	Exposure to war → delayed menarche (effect removed with the cessation of war)	[64]
Veronesi and Gueresi (1994)	Participants in a cross-sectional study	Italy	2930	Location of residence, father’s occupation, physical activity in years before menarche and family structure	Urban residence and lower occupational class of father → earlier menarche Higher levels of physical activity and first born → later menarche	[44]
Wronka et al. (2005)	Cross-sectional study of school girls	Poland	3271	Socioeconomic position (place of residence before high school, parental education and number of children in family)	Urban area of residence → earlier menarche Higher level of paternal education → earlier menarche Larger family size → later menarche (for those in urban and in an area with more people in agriculture than in industry)	[39]

Other studies have found that rapid prepubertal weight gain and childhood obesity are associated with an earlier menarche [33]. Using the growth status at birth as denoted by the expected birthweight ratio (EBW; a ratio of observed birthweight over median birthweight appropriate for the maternal age, weight, height, infant sex and gestational age), an Australian study found that earlier menarche was predicted by lower EBW combined with higher BMI during childhood [34]. Findings from the UK Newcastle One Thousand Families study support the notion of an interaction between birthweight and weight much later in childhood; specifically, girls who were large for gestational age and heavy at the age of 9 years had the earliest menarche [35]. From a life history perspective, results from the above studies appear to support the energetics theory for timing of menarche. While improvements in childhood nutritional status over time are thought to underlie the secular declines in age at menarche, the precise role of dietary composition in childhood remains unclear [36]. It is also worth noting that a recent study found no evidence to support a link between the secular rise in the prevalence of childhood obesity and the decline in average age at menarche [37].

Childhood socioeconomic position

Numerous studies have investigated the relationship between childhood socioeconomic characteristics (e.g., urbanization, education, parental occupational class and family size) and age at menarche (Table 1) [38-40]. In general, studies from both developed and developing countries have found that living in urban areas [40-44], having a father of higher occupational class [24,40,45-47] and having parents with higher educational levels [40,48] are associated with earlier menarche. Only a few studies have not found an association with area of residence [47,49]. One explanation postulated for the delayed menarche that is found among girls in rural areas is their increased levels of physical activity compared with those from urban areas [49]. Results from studies of the relationship between family size and pubertal timing have been inconsistent [50], with some finding a positive association[39,42,47,51] and others not [40,50].

Since socioeconomic circumstances often act as proxies for many factors, including the quality and quantity of food intake, energy expenditure, family structure and access to healthcare, few studies are able to separate out the effects of specific factors underlying the observed relationships between socioeconomic circumstances and timing of menarche. For instance, socioeconomic differences established in early childhood may be related to maternal efficiency during pregnancy (i.e., the ability of the mother to cope with and care for her unborn child), the duration of breastfeeding and the quality of the diet immediately after being weaned [48]. Associations between socioeconomic conditions and menarche may also be mediated by childhood growth; however, few studies have tested this hypothesis. Again, as socioeconomic circumstances influence nutritional environment in childhood, its relationship with the timing of menarche is consistent with the energetics theory of life history.

Psychosocial factors: childhood experiences

Role of family structure & relationships

While a mother’s age at menarche appears to be a better predictor of the daughter’s age at menarche than other external factors [52], research supports the role of family structure in determining pubertal timing. Paternal affection, positive family relationships and paternal involvement in child rearing are related to a comparatively later age of menarche [53,54], while increased family conflict, divorce and longer durations of paternal absence are correlated with earlier menarche [55-57]. It is also found that stepfathering predicts menarcheal age better than absence of a biological father and a longer presence of a stepfather correlates with earlier ages of menarche [58].

Exposure to external stressors/trauma

Results from the 1934-1944 Helsinki Birth Cohort have found that the 396 evacuees who were sent by their parents, unaccompanied, to temporary foster families in Sweden and Denmark because of the Soviet-Finnish wars, had earlier menarche than other girls [59]. Similarly, other studies examining the effect of migration have found increased occurrence of early sexual development, indicated through precocious puberty, in adopted children in cohort studies from European countries [60]. For example, a recent Danish study found that internationally adopted children were more likely (~15-20-times) to develop precocious puberty compared with the Danish reference group [61]. In addition, children migrating with their families had no increased risk of precocious puberty [61]. Exposure to other types of stressors in childhood, such as sexual abuse, have also been shown to be associated with the early onset of menarche [57,62]. These results concur with the studies associating childhood adversities, lack of paternal investment and father’s absence with earlier onset of menarche [22,53,63].

Studies have also found that age at menarche is delayed during periods of war. Authors working on the 1944-1945 Dutch Famine Study attributed this to the effect of food rationing [64]. In another study, it was found that girls living in besieged Srebrenica during the Balkan War had delayed age at menarche [65]. However, the author concluded that psychological trauma, physical injury and low socioeconomic conditions, as a result of war, may have been the causes. A study of girls living in the city of Sibenik, Croatia where the menarcheal status of girls was surveyed three-times in 1981, 1985 and 1996, found that there was a significant increase in mean menarcheal age of approximately 3 months between 1985 and 1996. This increase over time was dependent on the nature of the stress experienced with those girls who had experienced personal tragedies during the Balkan war, demonstrating an even greater delay in menarche of almost 11 months [50].

Research based on observational studies has also demonstrated delayed pubertal development following childhood adversity [50,64]. The effects of adversity may vary depending on the timing of exposure to the stressors [59]. For example, it has been shown that exposure to stress in early childhood is associated with earlier menarche [22], whereas exposure during or shortly before puberty has been associated with later onset of menarche [50,64].

Each of the three psychosocial models of pubertal timing within life history theory, namely psychosocial acceleration theory, parental investment theory and stress-suppression theory, can be invoked to explain the differing effects on age at menarche of each adverse condition in childhood identified previously. As an alternative, Boyce and Ellis have suggested the stress reactivity theory to account for both the delaying and accelerating effects of psychosocial stress on the timing of menarche [66]. They propose the notion that both highly protective and acutely stressful childhood environments trigger stress reactivity systems. If this triggering inhibits maturation of the hypothalamic–pituitary–gonodal axis, then this should produce a U-shaped relationship with age at menarche, whereby high social resources and support levels as well as high psychosocial stress and adversity are both correlated with later timing of menarche.

Age of menopause

Menopause marks the end of the reproductive phase of a woman’s life and usually occurs between the ages of 40 and 60 years [67-69], with the most reliable estimates indicating that the median age at menopause in Western industrialized countries is between 48 and 52 years [70]. While secular changes in average age at natural menopause are not as well documented as those in average age at menarche, there is some evidence to suggest that there have been modest increases in average age at menopause over the course of the 20th Century [71-74]. The impact of menopause is felt in physical, psychological and sociocultural domains and continues into the postmenopausal years.

Natural menopause is reached when a woman’s follicular reserve is depleted to approximately 1000 follicles, from a peak during fetal life of approximately 5 million at 20 gestational weeks, which has declined to approximately 2 million by birth [75,76]. The most important determinants of age at natural menopause are considered to be factors that affect the duration of decline in ovarian follicle reserve:

• Number of primordial germ cells that migrate to the gonadal ridge during intrauterine life;

• Mitotic abilities of these cells until gestational age of approximately 16-20 weeks;

• Rate of follicular atresia in both intra- and extra-uterine environments [76,77], suggesting that early life factors may play an important role in influencing age at menopause.

Throughout their reproductive lives, the general health and quality of lives of millions of women are affected by gynecological problems. Before reaching natural menopause, a significant proportion of women undergo treatment for such problems. The most radical of these treatments, namely hysterectomy and/or bilateral oophorectomy, end reproductive life. In these cases, women are said to have undergone surgical rather than natural menopause. Hysterectomy rates vary greatly between countries [78]. Our research from the MRC NSHD, a British cohort representative of women born in the 1940s, found that 22% of women had undergone hysterectomy by the age of 57 years with a mean age at hysterectomy of 43.6 years (range: 27-56 years). In this cohort, the most common reasons for undergoing hysterectomy were fibroids (30.5%) and menstrual problems (28.8%). Rates of hysterectomy in some other countries, including the USA and Australia, are higher than those reported in the UK [79,80]. For example, prevalence of hysterectomy was estimated to be between 33 and 43% among women aged 45-59 years in a recent study investigating women in the USA [81]. Despite these differences in prevalence between countries, reasons for performing hysterectomy are similar – US national health statistics suggest that between 1988 and 1993, fibroids were indicated in more than a third of hysterectomy cases [82]. As the processes that lead to the development of gynecological problems and that result in women seeking and undergoing treatment for such conditions are different to the processes that determine natural age at menopause, it may be expected that early life predictors of surgical menopause may differ from those of natural menopause.

Family and twin studies have indicated that the genetic effect on the timing of natural menopause is considerable, with estimates of heritability ranging from 30 to 85% [68,83]. Supporting this is evidence from cross-sectional and cohort studies that have demonstrated that a woman’s age at natural menopause is strongly associated with her mother’s reported age at menopause [68,84-89]. Studies also suggest that there is a genetic component to hysterectomy (i.e., surgical menopause) risk [90,91]. While the relationships between many adult environmental factors and timing of natural menopause have been investigated, only a few have consistently been related to an earlier menopause. Those that have are cigarette smoking [67,70,88] and nulliparity [90-94]. Some studies have also found a relationship between adult socioeconomic position and the timing of natural menopause with those women of lower position experiencing earlier menopause than women of higher position, even after adjustment for potential confounding factors, such as smoking and parity [95-98]. A range of factors in adulthood, including higher parity, lower socioeconomic position and greater changes in bodyweight over time, have been shown to be related to increased risk of hysterectomy [99-108].

As for age at menarche, a number of studies have examined the association between early life factors and age at natural menopause (Table 2) and these can be grouped under similar themes. Unlike the life history theory developed for the timing of menarche, and possibly owing to the range of influences operating during adulthood, no equivalent comprehensive framework has been developed to explain age at menopause. Fewer studies have examined early life factors in relation to hysterectomy risk, but where such studies have been undertaken their findings are described below.

Table 2.

Early life factors and their associations with age at natural menopause.

Study	Study population	Country of study	n	Early life factor(s) examined	Association(s) with age at menopause	Ref.
Cresswell et al. (1997)	Cohort study of women born in Hertfordshire and Sheffield	UK	755	Birthweight and weight at the age of 1 year	Higher birthweight → earlier menopause Higher weight at age 1 year → later menopause	[110]
Elias et al. (2003)	Women participating in a breast cancer screening program	The Netherlands	9471	Exposure to Dutch famine or food deprivation in Indonesia or Germany between 1944 and 1945	Severe or moderate exposure to famine → earlier menopause	[146]
Hardy and Kuh (2002)	Medical Research Council National Survey of Health and Development	UK	1514	Birthweight, breastfed, height and weight at the ages of 2 and 7 years, father’s occupational class at birth and household crowding at the age of 2 years	Higher weight at age 2 years and longer duration of breastfeeding → later menopause Socioeconomic position → no association	[111]
Hardy and Kuh (2005)	Medical Research Council National Survey of Health and Development	UK	1515	Father’s occupational class, household crowding, housing tenure and shared kitchen in childhood and parental divorce	Lower socioeconomic position → earlier menopause Parental divorce before a child’s age of 5 years → earlier menopause	[147]
Lawlor et al. (2003)	British Women’s Heart and Health Study	UK	3513	Childhood socioeconomic position (as indicated by father’s occupation, bathroom in house, hot water supply in house, shared bedroom, car access and age at which full-time education is completed)	Lower socioeconomic position → earlier menopause	[114]
Lumey et al (1997)	Women born in a specific hospital between 1944 and 1946	The Netherlands	700	Exposure to Dutch famine by trimester	Trimester of exposure to famine → no association	[143]
Mishra et al. (2007)	Medical Research Council National Survey of Health and Development	UK	1583	Breastfeeding, birthweight, bodyweight at the age of 2 years, cognition at the age of 8, 11 and 15 years, socioeconomic position at the age of 4 years and parental divorce by the age of 15 years	Never breastfed; lower cognitive ability and parental divorce → earlier menopause Heavier weight at 2 years → later menopause	[89]
Pesonen et al. (2008)	Helsinki Birth Cohort	Finland	511	Evacuation without parents during the war	No association	[59]
Strohsnitter et al. (2008)	Participants in the National Cooperative Diethylstilbestrol Adenosis Project	USA	4025	Maternal smoking during pregnancy	Exposure to cigarette smoke prenatally → earlier menopause	[148]
Treloar et al. (2000)	Prospective study of twin pairs	Australia	323 twin pairs	Birthweight	No association across the entire range of age at menopause Individual twins with very premature menopause were heavier at birth	[109]

Birthweight, childhood growth & early life nutrition

A number of epidemiological studies have investigated the influence of birthweight, childhood growth and early life nutrition on age at natural menopause. A twin study in Australia [109] and a cohort study of women in Hertfordshire, England [110], found no evidence that higher birthweight was associated with later menopause. However, low weight at 1 year was found to be associated with earlier menopause in the Hertfordshire cohort [110]. From the MRC NSHD study, it was found that women who had been breastfed had later menopause than other women [89,111]. Further evidence that early nutrition or growth might be important comes from a study of women in New Guinea [112], where the median age of menopause in a population who had suffered severe and prolonged malnourishment, and who were consequently of short height and low weight, was estimated to be 4 years earlier than women in the same region with better nourishment. This suggests that malnourishment (possibly acting prenatally through maternal undernutrition or postnatally through poor childhood growth) may play a role in early menopause.

A study of women exposed to the Dutch famine of 1944-1945 provides further support for the role of postnatal nutrition in influencing age at natural menopause – women who had been exposed to severe caloric restriction, especially those who were aged 2-6 years at the time of exposure, had an earlier natural menopause than those who were not exposed to these restrictions [113]. Famine exposure was also related to a higher occurrence of hysterectomy [113], with those women who were exposed to severe famine at increased risk of hysterectomy compared with those who were not (HR: 1.53; 95% CI: 1.27-1.84) [113]. The associations of birth size, childhood growth and early life nutrition with hysterectomy are less well researched. In the MRC NSHD where the association of weight in early childhood with hysterectomy was examined there was no evidence that weight in childhood was associated with subsequent hysterectomy risk [107].

Childhood socioeconomic position

In the MRC NSHD, early life socioeconomic position has been found to be more strongly associated with age at natural menopause than adult position. In another study of over 3000 British women, adverse socioeconomic circumstances in childhood, as well as in adulthood, were associated with age at natural menopause. In this study, it was found that the association between childhood deprivation and early menopause may be, at least in part, mediated via childhood diet, with this affecting both linear growth and age at menopause [114]. An association between childhood socioeconomic position and hysterectomy has also been found in the MRC NSHD with those women living in poorer socioeconomic conditions and with fathers of lower occupational classes at greater risk of hysterectomy than women living in better socioeconomic conditions and with fathers of higher occupational classes [106,115]. When similar associations were tested in two other British cohorts, it was found that the associations were acting in the opposite direction among women born in the 1920s [106]. This suggests that these are dynamic associations that change over time, possibly as a result of changes over time in a range of other factors, including access to medical care, attitudes of doctors, childbearing patterns and the availability of alternative treatments. Associations between childhood socioeconomic position and hysterectomy have not always been consistently found – in an American study no evidence of an association between various indicators of socioeconomic position in early life and hysterectomy risk was found among women enrolled in the Wisconsin Longitudinal Study [116].

Parental divorce during childhood

In previous analyses of the MRC NSHD, it was found that women who experienced parental divorce early in life had an earlier natural menopause than other women, raising the possibility that early emotional stress may also be a contributing factor [2,111]. Parental divorce before the age of 15 years not only exerts a strong influence on bringing forward natural menopause, but its effect is more than doubled for women with menopause before the age of 50 years compared with other women. It may be acting as a marker for early emotional stress and, in earlier findings, has been related to a range of psychological disturbances in childhood, such as bedwetting and delinquency [117,118], as well as poorer psychological health in adulthood [119]. The impact of stress responses both in early and later life, including responses of the hypothalamic-pituitary-adrenal axis, may influence age at menopause [120-123]. The life history theory of psychosocial acceleration as described with respect to the age at menarche may also be at play here. Another potential pathway that may be operating and, hence, explain these findings is the influence of psychological stress on the ovaries by lowering telomerase activity, a cellular enzyme that may act as a marker of ovarian functional age through its association with follicular atresia [124,125].

Interactions between early life factors & age at menopause

More recently, analyses of the MRC NSHD have revealed that some early life factors have a differential effect on age at natural menopause, which is dependent on whether women reach menopause before or after the age of 50 years [89]. The evidence indicates that weight at 2 years, mother’s reported age at menopause and parental divorce are more strongly associated with age at menopause if women have reached menopause before the age of 50 years compared with women who have reached menopause after the age of 50 years. In this study, a mother’s reported age at natural menopause may be considered as a simple proxy for the heritability of age at menopause, which in turn reflects genetic influences on the overall size of the follicle store before and shortly after birth and the rate of follicle atresia.

The differential effect according to age at menopause may be explained if the stress responses interact with another factor, for instance, the genetic setting of the size of the initial follicular reserve. Similarly, the time-dependent influence of weight during childhood on timing of natural menopause may indicate a postnatal interaction with factors that set ovarian function and development. On the other hand, the effects may still be caused by changes in environmental factors. For instance, weight during childhood may be less strongly related to stress than previously, due to the greater availability of food and lower levels of physical activity in current compared with previous generations. We are not aware of any studies that have examined the associations between parental divorce, psychological stress/trauma and hysterectomy.

Relationship between age at menarche & age at menopause

There is conflicting evidence concerning the relationship between ages at menarche and menopause. Some studies have found a relationship between earlier menarche and earlier natural menopause or perimenopause [92,94,126,127], a few have reported a relationship between earlier menarche and later menopause [128,129], but most others have found no association [69,130-133].

Conclusion & future perspective

Thomas et al. suggested that age at menarche is mainly determined by extrinsic factors, such as living conditions, while age at menopause appears to be influenced by intrinsic factors, such as reproductive history [14]. In this article, we have found consistent evidence to suggest that ages at menarche and menopause are both affected by a woman’s exposure to early life factors. The main early life factors associated with earlier menarche are:

• Higher growth rate during childhood

• Higher childhood socioeconomic position

• Family conflict and parental divorce

• Presence of stepfather

• Exposure to stressors during or shortly before menarcheal age

Those associated with earlier menopause are:

• Not having been breastfed

• Poor early growth

• Poor socioeconomic conditions

• Parental divorce

It is worth noting that while some early life factors – such as growth rate, childhood socio-economic conditions and parental divorce – affect both ages at menarche and menopause, the nature of these relationships differs. For instance, while parental divorce appears to bring about an earlier onset of both menarche and menopause, poor childhood growth reduces reproductive lifespan by delaying menarche and decreasing the age of menopause. Other risk factors also seemed to be unique to the age of menarche, such as the presence of stepfather and exposure to stressors during or shortly before menarcheal age.

Relatively little work has been done to investigate the extent to which childhood diet underlies the relationships between childhood growth and age at menarche or menopause, and further work is required. Future work to disentangle the effects of the different types of stressors in early life on pubertal timing is also required. The full impact of the rise in childhood obesity in the last few decades on the variability in ages at menarche and menopause is as yet unknown. More work is also required before any conclusions can be made about the associations of early life factors with risk of surgical menopause (i.e., hysterectomy) as these have been much less well studied than natural menopause. Furthermore, many of the results we have described need to be replicated in other studies before definitive conclusions about the early life predictors of menarche and menopause can be drawn. In this article, we have focused on menarche and menopause. However, to gain a full understanding of the associations between early life factors and reproductive health, there is also a need to consider the links of early life factors with other markers of reproductive health, such as fertility and gynecological disorders. These reproductive outcomes and their associations with early life factors have yet to be studied extensively and further research is required.

Beyond early life: continuity of reproductive health across the life course

Evidence of continuity in reproductive health across the life course indicates that for some women, the burden of poor reproductive health may be lifelong. For example, women who have early menarche are more likely to have menstrual problems [134]; those with menstrual problems are more likely to experience gynecological problems [135] and subfertility, which have been linked to earlier menopause [136]. A number of biological and social pathways may be operating across the life course that account for this continuity. These include variation in lifelong exposure or susceptibility to estrogen or other hormonal mechanisms, personality and psycho logical susceptibility and continuity in the socioeconomic environment. Each of these may influence and be influenced by body composition and diet. These findings suggest that the study of reproductive health would benefit from an integrated approach covering the whole life course, rather than each reproductive health outcome being studied in isolation.

From heritability studies, it has been shown that genetic factors play a key role in determining ages at menarche and menopause [20,21,68,83]. However, relatively little work has been done to unravel the complex interplay of genetic, social and behavioral factors (via epigenetic mechanisms), in explaining the variability in ages at menarche and menopause. One such approach would be to use family data and specific family-based studies (intergenerational, sibling and twin studies) across the life course to test specific causal mechanisms and life-course models since they can help in the understanding of whether the timing of risk factors (critical and sensitive periods) are important [137]. This analysis could be done in several ways, for instance, by comparing within and between sibling and twin associations for age at menarche where tight control for fixed factors within twin pairs and sibling groups is possible, and by comparing parental–offspring association to test causality for intrauterine exposure.

It is important to go beyond simple associations to understanding whether or not timing matters with respect to interventions to improve population health. For example, if an exposure during a developmental period in early life is causally related to age at menopause, but only through strong tracking of that exposure across the life course (e.g., bodyweight), then one would need to consider when it is easiest to most effectively alter the exposures. If it is easier to alter it in adulthood, then interventions in adulthood would be most appropriate [138]. With numerous family-based studies now underway, opportunities will arise to undertake such research in the future.

Executive summary.

Rationale for studying age at menarche & age at menopause

• Ages at menarche and menopause have been shown to be associated with adverse health outcomes in later life.

• Evidence of continuity in reproductive health across the life course indicates that for some women the burden of poor reproductive health may be lifelong, for example, women who have an earlier menarche are more likely to have menstrual problems and subfertility, which are linked to earlier menopause.

• The length of time between menarche and menopause determines the natural reproductive phase in a woman’s life.

Age of menarche

• Main early life factors associated with earlier menarche include:

  • –
    Higher growth rate during childhood
  • –
    Higher childhood socioeconomic position
  • –
    Family conflict and parental divorce
  • –
    Presence of stepfather
  • –
    Exposure to stressors during or shortly before menarcheal age

Age of menopause

• Main early life factors associated with earlier natural menopause include:

  • –
    Not having been breastfed
  • –
    Poor early growth
  • –
    Poor cognitive ability
  • –
    Poor socioeconomic conditions
  • –
    Parental divorce

Future perspective

• Research on the impact of childhood diet and the rise of childhood obesity on ages at menarche and menopause is currently lacking.

• More research is required on the effects of early life factors on the risk of surgical menopause.

• There is a need for further consideration of associations between early life factors and other markers of reproductive health, such as fertility and gynecological disorders.

• The study of reproductive health would benefit from an integrated approach covering the whole life course.

• Collecting family data or conducting a family-based study could help in testing causal mechanisms and in timing interventions.